Pilus biogenesis of Gram‐positive bacteria: Roles of sortases and implications for assembly

Abstract

Successful adherence, colonization, and survival of Gram‐positive bacteria require surface proteins, and multiprotein assemblies called pili. These surface appendages are attractive pharmacotherapeutic targets and understanding their assembly mechanisms is essential for identifying a new class of ‘anti‐infectives’ that do not elicit microbial resistance. Molecular details of the Gram‐negative pilus assembly are available indepth, but the Gram‐positive pilus biogenesis is still an emerging field and investigations continue to reveal novel insights into this process. Pilus biogenesis in Gram‐positive bacteria is a biphasic process that requires enzymes called pilus‐sortases for assembly and a housekeeping sortase for covalent attachment of the assembled pilus to the peptidoglycan cell wall. Emerging structural and functional data indicate that there are at least two groups of Gram‐positive pili, which require either the Class C sortase or Class B sortase in conjunction with LepA/SipA protein for major pilin polymerization. This observation suggests two distinct modes of sortase‐mediated pilus biogenesis in Gram‐positive bacteria. Here we review the structural and functional biology of the pilus‐sortases from select streptococcal pilus systems and their role in Gram‐positive pilus assembly.

Introduction

Gram‐positive bacterial infections are a significant global burden, and antibiotics remain the primary treatment of these infections.1, 2 Widespread bacterial resistance to available antibiotics has reinvigorated scientific interest in identifying novel anti‐infective and pathogen clearance therapies that target the virulence of pathogens without killing or inhibiting their growth.3 An understanding of the mechanisms underlying bacterial adhesion and colonization, host immune evasion, and antibiotic resistance is essential for the success of these efforts.4, 5 In this context, molecular entities, such as the pili that prime the bacteria for pathogenesis, have become the primary targets of many investigations.

More than one type of pili has been identified in Gram‐positive bacteria.6, 7 Covalent attachment of these fibrillary pili to the cell wall peptidoglycan is catalyzed by an enzyme called sortase.8 Sortases are cell surface‐associated or anchored transpeptidases that exhibit specificity toward proteins with a C‐terminal cell wall sorting signal (CWSS), which consists of an LPXTG motif followed by a hydrophobic domain and short tail of positively charged residues.9, 10 Most Gram‐positive bacteria encode multiple sortase enzymes with diverse sorting signal specificities (Fig. 1).11, 12 Understandably, both sortases and pili are targeted for pharmacotherapeutic development.13, 14

Figure 1.

Schematic representation of the pilus clusters and their components in some Gram‐positive bacteria. The major, basal, and tip pilin genes are represented in orange, green, and yellow colors; the genes encoding sortase enzymes are shown in different shades of pink, and the lepA/sipA is shown in blue. Other elements are represented in gray.

Since the discovery of sortases, our knowledge of Gram‐positive pilus biology has progressed rapidly, and excellent reviews describing the general principles of pilus assembly are available.15, 16, 17 In light of the recently determined crystal structures of many sortases and pilus components, in this review we present current knowledge of select Gram‐positive pilus clusters, their cognate sortases, elements that define the sortase enzyme substrate specificities, and external factors involved in pilus assembly. These insights may facilitate future therapeutic design efforts.

The Sorting Reaction

Staphylococcus aureus sortase A (Sau‐SrtA), which was first identified from a mutant strain defective in cell surface protein anchoring, is one of the best studied sortases.8 A multitude of surface proteins, including MSCRAMMs (microbial surface components recognizing adhesive matrix molecules), that play key roles in bacterial pathogenesis and infectivity are substrates of Sau‐SrtA. S. aureus srtA mutant strains are defective in their ability to establish infections, including infections associated with arthritis, endocarditis and lethal sepsis in mouse models of staphylococcal disease.18

Many cell wall‐anchored proteins in S. aureus and other Gram‐positive bacteria are synthesized as precursor proteins equipped with two signal motifs: an N‐terminal signal peptide19 for Sec‐dependent secretion20, 21 and a C‐terminal cell wall sorting signal (CWSS). The CWSS consists of a five‐residue sorting motif (consensus LPXTG sequence where X can be any residue) that precedes a hydrophobic and positively charged terminus.9, 22, 23 After translocation, the substrate protein is retained in the membrane via the hydrophobic and charged tail of the CWSS (Fig. 2). A sortase enzyme recognizes the sorting motif and cleaves between the Thr and Gly residues to form an acyl enzyme intermediate. A nucleophilic attack by the amine group of the pentaglycine from the S. aureus peptidoglycan precursor resolves the intermediate and releases the enzyme. The substrate product covalently linked to the peptidoglycan precursor is incorporated into the cell wall through the transglycosylation and transpeptidation reactions.

Figure 2.

Schematic of the sorting reaction in S. aureus sortase A. (1) The unfolded precursor substrate polypeptide is translocated across the cell membrane via the Sec translocon. (2) The folded protein is retained in the membrane by the terminal hydrophobic and charged regions, and its LPXTG motif is recognized by the nearby housekeeping sortase. (3) The amine group of the lysine of the pentaglycine cross‐peptide of lipid II attacks the resultant acyl‐enzyme intermediate. (4) Following the nucleophilic attack, the enzyme is released, and the substrate is covalently linked to lipid II. (5) The surface protein is incorporated in the peptidoglycan cell wall by transglycosylation and transpeptidation reactions.

Most Gram‐positive bacterial species contain multiple sortase enzymes.11 These enzymes were initially named based on either the order of their discovery or their chromosomal positioning, which caused problems and confusion due to a large number of sortases identified by many research groups in a short span of time. Emphasizing the need for a single unambiguous classification scheme, Comfort and Clubb12 and Dramsi et al.24 initially proposed four structural groups of sortases based on phylogenetic analyses and primary sequence similarities. Later, Spirig et al.25 and Bradshaw et al.26 defined two new groups, resulting in a scheme of six sortase classes (A–F), which is followed in this review. Each class recognizes a distinct sorting signal motif and variants thereof and performs mostly distinct functions.26

The class A sortase Sau‐SrtA is constitutively expressed; srtA is localized and regulated independent of its LPXTG motif‐containing substrates.10, 27 The canonical class B enzymes recognize a distinct NPQTN motif or its variants, and srtB is expressed and regulated within the isd locus. S. aureus SrtB (Sau‐SrtB), which is the class B prototype, has a distinctly different anchoring position on the peptidoglycan than Sau‐SrtA.28, 29 The class D sortase of B. anthracis (Ban‐SrtC) is required for spore formation and recognizes two substrates (BasI and BasH), both of which contain the LPNTA sorting motif.30 The Ban‐SrtC gene is colocalized with basI within an operon, whereas basH is located elsewhere in the genome. Interestingly, BasH is anchored to the forespore peptidoglycan, whereas BasI is anchored to the predivisional cell wall. Notably, modifications in the peptidoglycan architecture of bacterial species are not uncommon, and the cell wall anchor structures of classes A, B and D are characteristically distinct. The Cpe–SrtD gene from Clostridium perfringens and its LPQTG motif‐containing substrate are clustered, and its cell wall‐anchoring preference is not known.31 Similar to the members of class A, the class E sortases are not co‐localized with their substrates and recognize another noncanonical LAXTG motif.25, 26 The class C sortases constitute the second largest group of sortases and are involved in pilus assembly.25

Crystal structures of sortases from different classes reveal a conserved eight‐stranded β‐barrel core, whereas the connecting loops vary in length, conformation and embedded helices (Fig. 3).26 The invariable catalytic Cys, His and Arg dictate the catalytic efficiency, whereas the loops, especially the β6/β7, β7/β8, and β3/β4 loops, have been implicated in defining substrate specificity.32, 33, 34, 35 Notably, although the β6/β7 and β3/β4 loops are involved in recognition of the sorting motif, the β7/β8 loop has been implicated in binding of the peptidoglycan precursor for the transpeptidation reaction. Despite the extensive investigation of Sau‐SrtA, the mode of the catalytic reaction and the substrate recognition/binding motifs for sortases are controversial and remain unresolved due to the limited available structures of sortases with substrate analogs. Variations in the peptidoglycan cross‐linking preference across bacterial species and the ability of more than one class of sortase to anchor specific substrates to the cell wall have complicated our understanding of the transpeptidation reaction.

Figure 3.

Structures and topology of Class A–D sortase enzymes. (1) Sau‐SrtA (PDB code: 1T2P). (2) Sau‐SrtB (PDB code: 1NG5). (3) Spn‐SrtC1 (2W1J), and (4) Ban‐SrtD (PDB code: 2LN7).

Gram‐Positive Pilus Gene Cluster Components

Pilins

Pili were identified in Gram‐positive bacteria, including Actinomyces naeslundii,36 Bifidobacterium bifidum,37 Corynebacterium renale,38 Corynebacterium diphtheria,39 Enterococcus faecalis,40 Streptococcus agalactiae,41 Streptococcus pneumonia,42 Streptococcus pyogenes,43 Streptococcus suis,44 and many others.6, 45 Typically, pili are composed of at least two and often three individual multidomain proteins called pilins. Present knowledge of pilus assembly is based on the seminal work on C. diphtheriae, which expresses three distinct pili: SpaABC, SpaDEF, and SpaHIG (Spa for sortase‐mediated pilus assembly).39 The SpaABC pilus is a heteromolecular assembly of covalently interlinked SpaA units (major pilin), with SpaC and SpaB located at the distal (tip pilin) and proximal ends (basal pilin), respectively. SpaB is also interspersed along the fiber length.39 This arrangement is replicated in the SpaDEF and SpaHIG pili and other bacterial species with the exception of differences in the incorporation of the basal pilin.46 For instance, SpaE of the SpaDEF pili appears to be linked to SpaD in a manner distinct from the linkage between SpaA and SpaB of the SpaABC pili.46 Differences in the base pilin localization are also observed across species. For instance, the minor pilin RrgC of the TIGR4 pili of S. pneumoniae is located exclusively at the base,47 whereas the equivalent S. agalactiae pilin GBS150 of the PI‐2a cluster is distributed along the shaft48 similar to the corynebacterial pilus arrangement, suggesting that the distribution of the minor pili may not be a conserved feature. Pili devoid of basal pilins are expressed in A. naeslundii and B. cereus.49, 50 Of the four putative pilus gene clusters (srtBCD, srtE, srtF, and srtG) in S. suis,44 a deviation in the srtF cluster results in a pilus composed of only the major pilins mainly due to a mutation in the gene encoding the minor pilin,51 however, monomers of this minor pilin are anchored onto the cell surface.25

Many unique structural features distinguish the pilins of Gram‐positive bacteria. Most pilins have similar core structures with slight variations but exhibit limited primary sequence similarities. These pilins are modular proteins with frequent variant immunoglobulin folds of the CnaA and CnaB types and occasionally von Willebrand A (VWA)‐like and thioester domains.52, 53, 54 An isopeptide bond between the proximal Lys and Asn side chains catalyzed by a suitably situated Glu or Asp was identified in the pilin Spy0128.55 These stabilizing interactions56 were later confirmed in all Gram‐positive pilins and surface proteins with available crystal structures. Mutations that abrogate the formation of these bonds have been shown to affect the thermal stability of the pilins and their resistance to proteases.57 This rigidity and resistance are probably essential for withstanding the mechanical perturbations caused by the host58 toward bacterial purging. Two types of isopeptide bonds are seen; the D‐type connects two β‐sheets (sheet I and sheet II) when the Lys and Asn hosting strands are antiparallel and the essential catalytic Asp is present on a neighboring strand of sheet II, whereas the E‐type isopeptide bond connects two adjacent parallel strands of the same sheet (sheet I) when the conserved Glu is present on a proximal strand of sheet II. Importantly, isopeptide bonds confer considerable strength and rigidity to these surface adhesins, probably for with standing mechanical perturbation caused by the host toward bacterial purging. In addition to isopeptide bonds, an ester bond joining the Thr and Gln side chains was observed in a putative Clostridium perfringens surface protein exhibiting an IgG‐like fold; this bond provided mechanical strength to this surface adhesin.59 A widespread presence of similar ester bonds among other pili/adhesins can be expected, since many structural features are common between the pilins and surface adhesive proteins. Similarly, a thioester bond between the Cys and Gln implicated in covalent adhesion to pathogen cell surfaces by the mammalian complement proteins C3 and C4 was seen in the S. pyogenes tip pilin Spy0125 and later confirmed to be present in many Gram‐positive surface proteins.54 However, internal thioester bonds do not contribute to protein stability; instead, these bonds facilitate covalent interactions with host cell ligands.60

Pilus‐sortases

Class C sortases essential for pilus polymerization and assembly recognize a wide range of LPXTG variants. Here, we define pilus‐sortases as enzymes primarily involved in the polymerization and incorporation of individual pilins, and that may also perform secondary functions, such as anchoring pilins. All known class C sortases and a few class B sortases are pilus‐sortases; the genes that encode these enzymes and their substrates lie within a single cluster (Fig. 1). Examples with atypical characteristics include the class B pilus‐sortases of S. pyogenes and S. pneumoniae and the housekeeping sortase in C. diphtheriae, which is a class E enzyme.61, 62 Furthermore, Clostridium difficile sortase B belongs to the class B sortases but is not associated with heme or heme acquisition proteins; instead, this enzyme functions similar to a housekeeping sortase and recognizes an entirely different sorting motif (SPxTG or PPxTG).63

Additional factors

The complexity of pilus biogenesis arises from gene organization, the numbers and types of pilins, the numbers and specificity of pilus‐sortases and many known and unknown additional factors (Fig. 1). With few exceptions, pilus components are not constitutively expressed but instead are tightly regulated and often include transcriptional regulators.43, 64 Many clusters also express LepA/sipA, which have a signal peptidase‐like protein structure but are suggested to exhibit a chaperone‐like function.65 Pilus expression and assembly are complex processes that are often well‐regulated by factors that include transcriptional regulators encoded within clusters, heterogeneous expression, two‐component regulatory systems, and environmental factors, such as pH, temperature and oxygen availability;66, 67, 68, 69 these mechanisms are still being explored.

Streptococcal Pilus Clusters and Components

Group B streptococcus (GBS)

S. agalactiae (GBS) causes life‐threatening infections in neonates, and pregnant and nonpregnant adults.70 GBS encodes three pilus clusters in two separate loci.48, 64 The PI‐1 cluster encodes three pilins (GBS80, GBS52, and GBS104), two pilus‐sortases (GBS‐SrtC1–1 and GBS‐SrtC2–1) and an AraC type transcriptional regulator (Fig. 1).64 A third predicted sortase in PI‐1 does not function in pilus assembly.64 GBS80 forms the pilus backbone with GBS104 at the tip. GBS52 is present at the base of the pilus and is not anchored as a monomer on the bacterial surface.64 Variants of a second cluster (PI‐2a and PI‐2b) encode three pilins each (GBS59, GBS150, and GBS67 and SAN1518, SAN1516, and SAN1519, respectively). Additionally, PI‐2a encodes a RogB type transcriptional regulator, and PI‐2b is the only cluster that encodes LepA. The PI‐1 pilus‐sortases are class C sortases; although SAN1517 (GBS‐SrtC1–2b) in PI‐2b is also a class C sortase, SAN1515 (GBS‐SrtC2–2b) does not belong to any defined class.71

The GBS pili promote adherence and biofilm formation, and the pilins play varied roles.48 For instance, only PI‐2a confers the biofilm‐forming phenotype.72 Additionally, GBS1478 adheres to the brain endothelium, whereas GBS59 mediates invasion and internalization but not adherence.73 Pili elicit host immune responses,74 and most strains express one or two pilus clusters. Thus, a global pilus‐based GBS vaccine has a real possibility of providing maximum coverage for multiple serotypes.14, 75

Streptococcus pneumoniae

As facultative anaerobes carried in the nasopharynx, pneumococci are responsible for serious illnesses in young children, the elderly and HIV‐infected individuals.76 The rlrA pathogenicity islet, which is named after the encoded transcriptional regulator rofA‐like regulator77, 78 and its three reported variants,79 encode three rlr‐A regulated genes for the pilins RrgA, RrgB, and RrgC and three class C pilus‐sortases [Spn‐Srt(B–D) or Spn‐SrtC(1–3), respectively]. A second pilus islet (PI‐2) encodes two pilins (PitB and PitA), two sortases (Spn‐SrtG1 and Spn‐SrtG2) and SipA. Formation of the PitB backbone is catalyzed by Spn‐SrtG1, whereas PitA and SrtG2 are dispensable for PI‐2 pilus formation.79

The tip pilin RrgA confers adhesive specificity. RrgA variants display similar adhesive properties despite differences in the domains containing the VWA and RGD motifs.80 However, all three pilins elicit an immunogenic response in animal models.81

Group A Streptococcus (GAS)

S. pyogenes (GAS) is responsible for infections such as severe necrotizing fasciitis, acute rheumatic fever and glomerulonephritis.82 The GAS pili are associated with the FCT loci, which are named for the fibronectin‐binding, collagen‐binding proteins, and T‐antigens that they encode.83 FCT types 1–9 are grouped based on the gene content and organization. A single FCT locus linked with a particular T‐antigen is present in a strain associated with multiple M proteins, and all GAS isolates analyzed reveal the presence of one pilus cluster.43, 84

Two thioester bonds in the tip pilin Cpa of most FCT clusters mediate covalent linkage of the GAS pili to distinct host receptors.85, 86 The GAS pili have also been implicated in conferring tissue preferences.84, 87

Sortase‐Mediated Pilus Biogenesis: Current Model

The working model of pilus biogenesis derived from investigations in C. diphtheriae pili reveals that pilus biogenesis commences as a biphasic process on the exoplasmic side of the cytoplasmic membrane.88, 89 First, the pilus‐sortases covalently link pilins to form a pilus. In the second step, the resultant fiber is anchored covalently on the cell wall by a housekeeping sortase [Fig. 4(A)].90 In addition to the C‐terminal sorting motif, the major pilin SpaA of the SpaABC pili hosts a conserved pilin motif with an essential Lys residue (WxxxVxVYPKN) and another glutamic acid‐containing motif called the E‐box (YxLxETxAPxGY); these motifs are necessary for pilin polymerization and minor pilin incorporation, respectively.39, 91 The pilus‐sortase Cdi‐SrtA, which is specific for the SpaABC pilus, forms acyl‐enzyme intermediates after cleaving between the Thr and Gly of the LPLTG motif of SpaA and SpaC. A covalently linked SpaC‐SpaA‐Cdi‐SrtA complex is generated following nucleophilic attack by the conserved Lys of the SpaA pilin motif; subsequent SpaA units are similarly incorporated head‐to‐tail to form an elongated fiber, which grows at the base.92, 93 Since SpaC lacks the pilin motif, this pilin is found at the tip of the fiber; SpaC is presumed to be the nucleating pilin90 for the addition of the SpaA pilins instead of an initiation factor for polymerization. The SpaA availability determines the length of the pilus in combination with the insertion of SpaB.88, 93, 94 K139 of SpaB is critical for its dispersion along the shaft via the SpaB‐Cdi‐SrtA complex and for the termination of polymerization and cell wall anchoring via the SpaB‐Cdi‐SrtF intermediate. The basic patch of the C‐terminal transmembrane domain of Cdi‐SrtA is critical for pilus polymerization,92 which suggests that its topological presentation in the membrane is vital for efficient pilus polymerization. In addition to topology, the regulation, expression, local concentrations and stoichiometry of the housekeeping and pilus‐sortases influence the efficiency and rate‐limiting steps of pilus biogenesis and anchoring.

Figure 4.

(a) Model of pilus assembly depicted for GBS PI‐1. (1) The Class C pilus sortases form acyl‐enzyme intermediates with GBS104 and GBS80. (2) Enzyme intermediates are resolved by the lysine in the pilin motif of GBS80, linking the two proteins with GBS104 at the tip while incorporation of GBS52 along the pilus is mediated by GBS‐SrtC1–1. (3) Repetition of the process interlinks GBS80 units in a head‐to‐tail manner and incorporates GBS52 along the shaft. In the second phase, presentation of GBS52 by the housekeeping sortase terminates assembly. (4) The complex between the assembled pilus and the housekeeping sortase is resolved by the nucleophilic attack from lipid II. (5) The assembled pilus is covalently linked and incorporated in the peptidoglycan by subsequent processes. (b) Schematic of S. pyogenes FCT‐2 assembly. Distinctly from the Group I biogenesis, here the Class B sortase, Spy0129, forms acyl‐enzyme intermediates with Spy0128 and requires the participation of signal peptidase‐like protein, SipA. Specific details of incorporation of the minor pilins and factors involved therein to form and anchor the assembled pilus on the cell wall are less clearly understood. Mechanistic details from the emergence of the folded pilins from the Sec translocon to recognition by the sortase, while still unclear, may or may not follow the same trajectory in Groups I and II models.

Biphasic pilus biogenesis, in which pilus assembly by a pilus‐sortase is followed by pilus anchoring on the cell wall by the housekeeping sortase, has been reported for many Gram‐positive bacteria.49, 90, 95 Housekeeping sortases act exclusively on the minor pilin and this is a requirement for the cell wall covalent attachment of the pilus.96 Untangling the mechanistic details of assembly for a particular cluster is complicated due to the presence of more than one pilus‐sortase in some clusters and variations across species. Unlike the SpaABC cluster, the SpaDEF and SpaHIG clusters encode two pilus‐sortases each. Cdi‐SrtB and Cdi‐SrtC of the SpaDEF pilus both recognize the LPMTG motif of SpaD; however, Cdi‐SrtB is essential for incorporating SpaE (sorting motif LALTG).46 Similarly, in GBS PI‐1, the sortases GBS‐SrtC1–1 and GBS‐SrtC2–1 share functional redundancy as the pilin polymerase for GBS80 (IPNTG) but are specific for the incorporation of the minor pilins GBS52 (IPKTG) and GBS104 (FPKTG), respectively.64 Polymerization of GBS59 (IPQTG) in PI‐2a can be catalyzed by either of the two pilus‐sortases; however, SAG1406 (GBS‐SrtC3–2a) is essential and specific for the incorporation of the base pilin GBS150 (LPKTG), whereas deletion of SAG1405 results in reduced incorporation of the tip pilin GBS67 (IPMTG).64 In a more complex scenario, the rlrA islet encodes three pilus‐sortases [Spn‐SrtC(1–3)] that are differentially involved in the incorporation of the three pilins (RrgB, RrgA, and RrgC, with sorting motifs IPQTG, YPRTG, and VPDTG, respectively), all of which lack the canonical Leu in the first position of the LPXTG motif. Falker et al.97 reported that Spn‐SrtC1 and Spn‐SrtC2 were redundant in their ability to polymerize RrgB. Some studies have attempted to clarify the roles of the accessory pilus‐sortases.98, 99 Spn‐SrtC2 links RrgA to RrgB but cannot incorporate RrgC,97, 99 whereas Spn‐SrtC3 is not a pilin polymerase and is associated with the focal presentation of the pilus on the bacterial surface.97 Notably, in contrast to the biphasic mode of pilus assembly, the pneumococcal pilus is not anchored to the cell wall by the housekeeping Spn‐SrtA but instead is anchored by pilus‐sortases, as reported by Lemieux et al.98 Recently, Shaik et al. showed that RrgC was the cell wall anchor attached to the peptidoglycan by Spn‐SrtA, and Spn‐SrtC3 was implicated in anchoring the minor pilins to the cell wall.100, 101 The S. pyogenes FCT regions encode the backbone protein FctA and one or two minor pilins in addition to the pilus‐specific sortases. Interestingly, a putative signal peptidase‐like protein called SipA is essential for pilus formation in serotype M3/FCT‐3, which has two minor pilins (Cpa and FctB).65 Spy‐SrtC2 is required to attach Cpa to the major pilin FctA. Since the pilin motif containing the essential Lys is not present in Cpa, this pilin is predicted to be found at the pilus tip [Fig. 4(B)].102 However, Cpa was shown to be located at the pilus base in the M49 serotype, which complicated the GAS pilus assembly story and its sequence of operations.103

Not all Pili are Created Equal: Structural Insights

Structural characterization of streptococcal pilus clusters is an emerging field of research. The successful structures include pilins, pilus‐sortases, and peptidases (Table 1). Currently, the structures of the major players in pilus assembly from more than one cluster and bacterial species provide a better picture for understanding the molecular basis and mechanistic details of pilus biogenesis.

Table 1.

Structures of Pilus Cluster Components from some Streptococcal Species

Species	Pilus cluster	Pilin/sortase	Sorting motif	PDB code	References
S. agalactiae	PI‐1	GBS80	IPNTG	3PG2, 3PF2	121
S. agalactiae	PI‐1	GBS52	IPKTG	3PHS	122
S. agalactiae	PI‐1	GBS104	FPKTG	3TVY, 3TXA, 3TW0	123
S. agalactiae	PI‐1	GBS‐SrtC1–1	–	4GIJ, 3TB7, 3RBI, 3RBJ, 3RBK, 3TBE	104, 106, 109
S. agalactiae	PI‐1	GBS‐SrtC2–1	–	4GIH	106
S. agalactiae	PI‐1	Sortase A	–	3RCC	104
S. agalactiae	PI‐2a	BP‐2a	IPQTG	2XTL	14
S. agalactiae	PI‐2a	GBS‐SrtC1–2a	–	3O0P	108
S. agalactiae	PI‐2b	BP‐2b	LPSTG	4UZG	124
S. agalactiae	PI‐2b	GBS‐SrtC1–2b	–	4D7W	71
S. pneumoniae	RlrA	RrgB	IPQTG	3RPK, 2Y1V, 2X9Y, 2X9W, 2X9X, 2X9Z	125, 126, 127
S. pneumoniae	RlrA	RrgC	VPDTG	4OQ1	100
S. pneumoniae	RlrA	RrgA	YPRTG	2WW8	53
S. pneumoniae	RlrA	Spn‐SrtC1	–	2WTS*, 2W1J	105
S. pneumoniae	RlrA	Spn‐SrtC2	–	3G66, 3G69	101
S. pneumoniae	RlrA	Spn‐SrtC3	–	2W1K	128
S. pneumoniae	PI‐2	PitB	VTPTG	4S3L	62
S. pneumoniae	PI‐2	Spn‐SrtG1	–	4Y4Q	62
S. pyogenes	FCT‐1	Spy0128	EVPTG	3B2M, 3GLD, 3GLE	55, 57
S. pyogenes	FCT‐3	FctB	LPLAG	3KLQ	129
S. pyogenes	FCT‐2	Spy0125	VVPTG	4BUG, 2XID, 2XI9, 2XIC, 4C0Z	60, 85, 87
S. pyogenes	FCT‐2	Spy0129	–	3PSQ	61
S. pyogenes	FCT‐3	SipA	–	4N31	130
S. pyogenes	FCT‐2	Sortase A	–	3FN5, 3FN6, 3FN7	131

Group I: Pilus biogenesis using class C pilus‐sortases

Class C pilus‐sortases display the conserved beta‐barrel fold of Sau‐SrtA and the presence of a ‘lid,’ which is a loop linking the last helix in the N‐terminal helical domain to the beta‐barrel core of the enzyme via a ‘hinge’ region; this ‘lid’, which is lodged in the putative active site, is their structural hallmark. The ‘lid’ is characteristically flexible and displays incomplete electron density definition and high‐temperature factors in most crystal structures. The Spn‐SrtC1 structure is an exception in which the electron density for the entire ‘lid’ is well defined. However, the Spn‐SrtC1 ‘lid’ also exhibits high‐temperature factors compared to the protein core, indicating flexibility of the ‘lid’. Asp and the hydrophobic residue of the DP(Y/W/F) motif make direct contacts with the catalytic Arg and Cys, respectively, and anchor the ‘lid’ in the binding pocket [Figs. 5(A) and 6(A)]. Additionally, the side chain of a hydrophobic residue located further down the ‘lid’ points into a hydrophobic patch present in the active site groove underneath.104 The conserved DP(Y/W/F) motif of the ‘lid’ tends to have a variable residue at the third position, whereas the Asp, which forms a salt‐bridge with the catalytic Arg, is an invariant residue. The interaction of the DP(Y/W/F) motif with the catalytic residues is an invariant in the class C pilus‐sortase structures. Mutation of the catalytic His or Arg residues rendered Spn‐SrtC1 inefficient in catalyzing the polymerization of RrgB in vitro.105 Additionally, mutants of the Asp or the third residue of the DP(Y/W/F) motif are less stable in thermal shift experiments, suggesting that the ‘lid’ confers protein stability and involved in the polymerization, most likely via recognition of the substrate sorting motifs.105 Similarly, a recognition or regulatory role for the ‘lid’ has been revealed for GBS‐SrtC1–1 and GBS‐SrtC1–2a.106, 107 However, the C‐terminal trans‐membrane region in the GBS PI‐1 and PI‐2a sortases is required for catalysis and pilin polymerization,108 Similar importance for the terminal regions was reported in C. diphtheria.92

Figure 6.

(a) Comparison of active site residues and ‘lid’ loops of the Group I pilus‐sortases: GBS‐SrtC1–1 (yellow, PDB code 3RBK) and Spn‐SrtC1 (green, PDB ID 2W1J and labeled in parenthesis). (b) Superposition of Class B pilus‐sortases, Spy0129 (green, PDB ID: 3PSQ) and Spn‐SrtG1 (magenta, PDB ID: 4Y4Q) reveal the presence of polar residues in the active site (S219 and D109 in Spy0129, and S239 and D130 in Spn‐SrtG1) compared to the hydrophobic residues (L182 and L103 in GBS‐SrtC1–1) in Class C pilus‐sortases.

However, because there is no evidence to suggest that the substrate can access the putative active site of a class C sortase in the ‘closed’ conformation, a ‘lid’ movement is probably required for substrate binding.104, 109 Interestingly, the catalytic triad remains accessible to small molecule inhibitors with the ‘lid’ in place,104 and recent NMR and molecular dynamics (MD) studies of Spn‐SrtC1 have underscored the energetic barrier involved in dislodging the ‘lid’.110 Since the ‘lid’ is proposed to be a pseudo‐substrate, LPXTG mimic peptides have been shown to stabilize ‘lid’ mutants105 by having favorable binding energetics.

Crystal structures with an ‘open’ probably active conformation are available for GBS‐SrtC1–1 and S. suis Ssu‐SrtC1 and provide a snapshot of ‘lid’ displacement and stabilization.111 In the GBS‐SrtC1–1 crystal structure, a substrate‐like loop from a symmetry‐mate was observed protruding into the active site of a neighboring molecule and displacing its ‘lid,’ which re‐conformed into a five turn α‐helix away from the enzyme active site.109 These structures provide a glimpse into possible ‘substrate‐induced’ conformational changes from the resting state to the activated enzyme [Fig. 5(B)]. Thus, the ‘lid’ possibly acts as a gateway for substrate recognition and binding. This serendipitous observation in GBS‐SrtC1–1 is supported by the Ssu‐SrtC1 structure, which is not burdened by any perceived crystallographic artifacts. Remarkably, superposition of the two structures reveals spatial conservation of the extra helix (i.e., the ‘lid’ in the ‘closed’ structures) [Fig. 5(C)]. Point mutations in the Spn‐SrtC1 ‘lid’ result in increased accessibility of substrate peptides and enhanced enzyme hydrolytic activity, thereby confirming the role of the ‘lid’ as a gateway to the active site. Additionally, the displaced ‘lid’ acquires the helical conformation suggested by NMR and MD simulation studies on point mutants.110

Figure 5.

(a) Structures of the GBS‐SrtC1–1 (cyan, PDB ID: 3RBK) and Spn‐SrtC1 (magenta, PDB ID: 2W1J), with the ‘lid’ loops (yellow and green, respectively) positioned in the active site. (b) Superposed structures of GBS‐SrtC1–1 in putative ‘closed’ (cyan, PDB ID: 3RBK) and ‘open’ (blue, PDB ID: 3TB7) conformations. The ‘lid’ is depicted in yellow color, positioned as a loop in the active site in the ‘closed’ conformation and a helix in ‘open’ conformations of GBS‐SrtC1–1. (c) Superposed structures of putative ‘open’ GBS‐SrtC1–1 (blue, PDB ID:) and Ssu‐SrtC1 (orange, PDB ID: 3RE9) with the displaced GBS‐SrtC1–1 ‘lid’ stabilized as a helix (yellow) spatially conserved with Ssu‐SrtC1 helix (magenta).

Elements that define the substrate specificity of the pilus‐sortases are not well understood. GBS‐SrtC1–1 and GBS‐SrtC2–2 can both polymerize GBS80, which has a sorting motif with an Ile residue in the first position. However, these pilus‐sortases are stringent in recognizing minor pilins that possess either Ile of Phe in the first position; the only difference between these substrates is their sorting motifs. This finding suggests that elements beyond the sorting motif clearly play significant roles in recognition and specificity. In the corynebacterial SpaABC pili, the specificity of the housekeeping sortase for the basal pilin was reported to reside within its sorting motif.112 A possibly similar situation in GBS would imply that GBS‐SrtA was equally stringent for residues in the first and third positions of the sorting motif of the basal pilin, since GBS‐SrtC1–1 recognized both IPNTG of GBS80 and IPKTG of GBS52. GBS52 has been reported to be present intermittently along the pilus shaft based on immunogold electron microscopy using antibodies against GBS52.48 However, a higher resolution study, as reported for S. pneumoniae, is needed to eliminate the possibility that the observed GBS52 along the pilus shaft is a result of crowding from multiple pili.47 Additionally, since the basal pilin triggers anchoring of the pilus by the housekeeping sortase, questions remain concerning why only a certain GBS52 unit at the base of a long fiber is processed for anchoring and not the first GBS52 incorporated along the pilus shaft. In addition to sortase specificity, sortase accessibility, and availability at the assembly line must play roles in modulating this fine balance. Importantly, class C and A sortases also differ in their membrane topology because class C enzymes possess both N‐ and C‐terminal transmembrane domains in contrast to the N‐terminal hydrophobic region of the class A sortases. Crystal structures of full‐length class C pilus‐sortases (or for that matter any sortase) in complex with full‐length substrates would enhance our understanding of the basis of their specificity. However, with the advent of major advances in cryo‐electron microscopy, the potential for visualization of these complexes and the detailed architecture of the pilus is especially exciting. The recognition and binding of large substrates also involve the conformational flexibility of the full‐length substrate protein compared to a short peptide motif.113 Hence, structure determination of these complexes using traditional methods is a tall order.

Group II: Pilus biogenesis requiring class B pilus‐sortases and LepA/SipA

Genetic, biochemical and structural data highlight the differences and distinctions among the pilus clusters. For instance, GBS PI‐1 and S. pyogenes FCT‐6,84 as well as S. pneumoniae PI‐2 and the S. pyogenes pili, share some features.79 The latter two clusters encode two sortases each, one of which (Spy0129 and Spn‐SrtG1) acts as a polymerase. Spy0129 and Spn‐SrtG1 are class B pilus‐sortases that recognize distinct sorting motifs: EVPTG in Spy0128 (or VPPTG in Cpa) and VTPTG of PitB from PI‐2, respectively. In contrast to Group I, neither pilus can be generated in vitro by mixing the pilin and polymerase, nor does SipA expression enable in vitro pilin polymerization. The in vivo polymerization efficiency seems to require expression of all components from the same operon.62

Although they retain the conserved sortase fold, Spy0129 and Spn‐SrtG1 differ from the class C sortases by exhibiting an ‘open’ and accessible substrate‐binding pocket with a greater electrostatic charge distribution. Class C sortases employ two hydrophobic residues (Leu103 and Leu182 in GBS‐SrtC1–1) in the substrate‐binding pocket that are replaced with Asp and Ser in both Spy0129 and Spn‐SrtG1, which is consistent with the lack of a shielding ‘lid’ in the binding pocket [Fig. 6(B)].104 Either of the first two positions of EVPTG and VTPTG contains a polar residue, whereas the sorting motif for Group I has an invariant Pro in the second position following a hydrophobic residue. Hence, the nature and interactions essential for recognition and binding between the two groups are different, and polar interactions may play a more significant role in Group II pilus‐sortases. The major Group II pilins are smaller in structure and are composed of two CnaB‐type domains, which is in contrast to the multidomain major pilins in Group I. K161 is the pilin‐conjugating lysine in Spy0128, and K203 of PitB superimposes with K161 of Spy0128; however, the identity of the critical lysine in PitB needs to be established biochemically. The structural landscape surrounding K161 in Spy0128 and K203 in PitB is similar, whereas K183 of RrgB in Group I resides on a much shorter loop connecting two short beta‐strands, indicating its limited maneuverability.

The common structural fold and divergent functions of class B enzymes in different bacterial systems raise intriguing questions about their acquisition, evolution, and substrate specificities. S. aureus and C. difficile class B sortases (Sau‐SrtB and Cdf‐SrtB) are not pilus‐sortases; whereas Sau‐SrtB is required for heme‐iron acquisition, Cdf‐SrtB functions as a housekeeping sortase. These enzymes contain the STC and YXH motifs at the catalytic Cys and His compared to the LTC and TAH motifs present in the class C sortases.34 These subtle differences should play a role in specificity because similar residues in Sau‐SrtA have been correlated with specificity.114, 115 The catalytic Cys and His of Spy0129 and Spn‐SrtG1 are localized at the edge of strand β7 and the beginning of a helix following strand β4, respectively, which is in contrast to the residues of Sau‐SrtB and Cdf‐SrtB. The analogous residues are positioned on strand β7 and a loop following strand β4 in many class C sortases in the ‘closed’ state as well as in Ssu‐SrtC1. In Spy0129 and Spn‐SrtG1, His is part of a YXHH motif with two consecutive His residues that is not observed in either Sau‐SrtB or Cdf‐SrtB. Furthermore, the β6/β7 loop of class B pilus‐sortases is more elaborate and includes a short beta‐strand (Fig. 7). Mutant Sau‐SrtA containing the β6/β7 loop of Sau‐SrtB can recognize and cleave the sorting substrate of Sau‐SrtB, and specific hydrophobic residues on this loop appear to facilitate substrate recognition by interacting with the Leu of LPXTG. Additionally, closure of the β6/β7 loop in response to calcium ions revealed an ‘open’ and ‘closed’ form in Sau‐SrtA.116 Lid‐swapping experiments in class C sortases demonstrated the role of the ‘lid’ in substrate recognition.105 In the absence of a ‘lid’ in the class B pilus‐sortases, the conformational states and significance of the β6/β7 loop in substrate accessibility and recognition remain to be determined. Notably, the class B pilus‐sortases and Cdf‐SrtB have a Gly in the fifth position of the sorting motif similar to Sau‐SrtA, whereas only Sau‐SrtB and Cdf‐SrtB recognize the substrate with a Pro in the second position. Positional preferences in the sorting motif, the distinct conformation of the β6/β7 loop and the more flexible disposition of the pilin‐conjugating Lys in the major pilins in Group II may all contribute to the selection and fine‐tuning of interactions with suitable substrates.

Figure 7.

Comparison of structures of Class B sortases from (A) S. aureus (green, PDB ID: 1NG5), (B) S. pneumoniae (pink, PDB ID: 4Y4Q), and (C) C. difficile (teal, PDB ID: 4UX7) with the β6/β7 loop highlighted in blue.

A signal peptidase‐like protein called SipA or LepA identified in GAS is present in Group II clusters. However, its function remains unclear. SipA2 of GAS serotype M3 is required for pilin polymerization with pilin T3 and sortase SrtC2 and links the ancillary pilins to the pilus shaft. Interestingly, SipA2 closely resembles the catalytic domain and peptide‐binding cleft of the E. coli SPase. The catalytic Ser and Lys residues, which are essential for the peptidase function, are replaced by Asp and Gly, although their mutation has no effect on pilus polymerization.117 Since SipA is essential for polymerization but incapable of functioning as a signal peptidase, SipA may function as a chaperone in T3 pilus formation. However, in Actinomyces oris, the SipA homologous protein Lepb2 but not Lepb1 is essential for pilus formation and requires the Ser‐Lys dyad for its signal peptidase function for the assembly of both type 1 and type 2 fimbriae.118

Pilus clusters with overlapping and novel features

GBS PI‐2b encodes LepA/SipA, three pilins and two pilus‐sortases (GBS‐SrtC1–2b and GBS‐Srt2–2b). Polymerization of the PI‐2b pilus is exclusively performed by GBS‐SrtC1–2b. Deletion of the gene encoding GBS‐SrtC2–2b produces pilus polymers that are released into the culture medium, suggesting that GBS‐Srt2–2b acts as a housekeeping sortase. GBS‐SrtC1–2b exhibits a unique core β‐barrel of 10 β‐strands, which is two more than GBS‐SrtC1–1 and GBS‐SrtC2–1,104, 106 and extra features, such as a β4/β5 loop with an extra beta‐hairpin. Interestingly, the only other Group I sortase structure that exhibits these two features is the ‘open’ state of GBS‐SrtC1–1 (PDB 3TB7), which suggests possible functional relevance. GBS‐SrtC1–2b is a class C pilus‐sortase with a ‘lid’ that has an MKW motif instead of the canonical DP(Y/W/F) motif. The ‘lid,’ while situated spatially at the same position as the third helix in the ‘open’ state of GBS‐SrtC1–1, could be an adventitious consequence of crystal packing. This ‘lid’ conformation was speculated to represent an intermediate state between the ‘open’ and ‘closed’ states of GBS‐SrtC1–1, especially given its low B factors and complete electron density definition.71 NMR investigations eliminated the possibility of multiple ‘lid’ conformations in Spn‐SrtC1,110 but the flexibility of the ‘lid’ in other pilus‐sortases on slow and fast dynamic timescales is not known. Interestingly, the side chain of the Trp residue of the MKW segment of GBS‐SrtC1–2b spatially superimposes with the side chain of the hydrophobic lid‐anchor residue of GBS‐SrtC1–1, which along with the DP(Y/W/F) motif is vital for locking the ‘lid’ in position. Ascertaining the conformation of the GBS‐SrtC1–2b ‘lid’ upon mutation of the Trp would be interesting, since GBS‐SrtC1–2b lacks the Pro of the DP(YW/F) motif that demarcates the third helix in GBS‐SrtC1–1 and Ssu‐SrtC1. The GBS PI‐2b cluster also encodes a signal peptidase‐like protein, as observed in Group II; however, this cluster differs because it encodes GBS‐SrtC1–2b, which is a class C sortase with a displaceable ‘lid.’ This cluster also encodes GBS‐SrtC2–2b, which is not a pilin polymerase and acts as a housekeeping sortase, but its primary sequence does not resemble any other prototypical sortase.71 Therefore, overlapping features in clusters are not uncommon; for example, the type I cluster of A. oris encodes the prepilin peptidase‐like FimR while expressing a canonical class C sortase with a ‘lid’ anchored in the active site.119 Furthermore, GBS‐SrtC2–2b lacks a ‘lid’, is specific for the minor pilin and catalyzes the anchoring of the pilus on to the cell wall. In this respect, GBS‐SrtC2–2b is similar to the FCT cluster of the S. pyogenes M3 cluster, where Spy‐SrtC2 anchors the minor pilin onto the cell wall instead of the housekeeping sortase.120 Spy‐SrtC2 specifically recognizes the sorting motif QVPTG, and mutations of this motif into the housekeeping LPXTG‐type motif abolish anchoring; this result suggests that the specificity of anchoring by the pilus‐sortase resides with the sorting motif sequence, at least in this system.

Conclusions

Despite the lack of 3D structures of full‐length pilin and pilus‐sortase complexes, the nearly complete structural characterization of many streptococcal pilus clusters ushers an exciting era that will bridge the gap in our understanding of the structural mechanisms integral to Gram‐positive pilus biology. Research on multiple pilus clusters from strains and serotypes of Gram‐positive bacteria has provided insights into both the commonalities and differences between clusters and the encoded pili. Analysis of the differences in the structural features and functionality of the pilus‐sortases has revealed two parallel themes that we speculate involve two distinct mechanistic pilus biogenesis pathways in Gram‐positive bacteria; these pathways need to be defined to elucidate the mechanistic steps of assembly via either pathway. These pathways pertain to clusters that encode at least one, but often more than one, class C pilus‐sortase that possesses a displaceable ‘lid’ in the putative active site (Group I, e.g., GBS PI‐1 and S. pneumoniae rlrA) as well as clusters that encode and utilize a class B sortase for pilus assembly in addition to a LepA/SipA peptidase‐like protein that is essential for pilin polymerization (Group II, FCT, and S. pneumoniae PI‐2). Based on structural data on class C pilus‐sortases, ‘lid’ displacement precedes or is concomitant with substrate binding and possibly involves a conformational transition to a helix formation that stabilizes the flexible ‘lid’; however, the mechanistic structural details remain elusive. Conversely, the substrate recognition and specificity of class B pilus‐sortases are completely unexplored. This lack presents an intriguing question in itself since some class B sortases may function as housekeeping sortases (e.g., C. difficile sortase B),63 whereas others function as pilus polymerases (Spy0129) or play roles in iron acquisition (Sau‐SrtB). Furthermore, the functions of the ‘lid’ and the signal peptidase‐like protein SipA must be understood in the overall mechanistic scheme of pilus biogenesis, including how (if at all) the structural and functional contributions of each factor are compensated in the other pathway. The multiplicity of pilus clusters may be linked to tissue tropism and species‐specific roles in pathogenesis. However, the need to divert bacterial resources for distinct mechanistic pilus production pathways is an intriguing question. Since duplication, redundancy, and versatility of surface adhesins help pathogens adapt to and escape the host response; multiple pilus types could also ensure survival and pathogenesis.